New Year's Eve in Fairbanks

How cold will it be when the ball drops on New Year's Eve in New York City is, for better or worse, in the news, so I (Rick T.) thought it might be interesting to look at New Year's Eve temperatures in Fairbanks.

With a little work I was able to pull together the temperature near midnight each December 31st since 1941 (thanks to Brian B. for a last minute data access rescue). These are the temperatures from the the Weather Bureau Cushman Street office for 1941, Ladd Field 1942, Weeks Field 1943-50 and Fairbanks International since New Years Eve 1951. I've used the observation closest to midnight Alaska Standard Time, which has typically been between 1150 and 1155pm, though a couple years in the World War Two era the closet observation was about 1130pm. I've used the standard time zone that was in effect at the time, so there is the one hour change in 1983. Hopefully then, this reflects the temperature (at the official observing site) at the time when Fairbanksians were toasting in the New Year. So here's what we get:

The first thing I notice is the 81F range of temperatures: the warmest New Year's Eve temperature was +28F in 2014, the coldest -53F in 1968, and the standard deviation is a whopping 20F. The second thing that stands out is that there seems to be a lot of New Year's Eve with temperatures near to slightly above 0F. This may be something of a surprise, given that the average temperature (in the 1981-2010 normals) for Dec 31 is -7F. In fact, the distribution of New Year's Eve temperatures (below) is typical of mid-winter temperatures in Interior Alaska, with a marked "left skew". That is, a long tail to the cold side, and a short tail to the warm side. This makes physical sense: in the absence of solar heating,  there are a lot of ways for an inland valley at 65N to be cold but very few ways to be really mild. I will not get into the details here, but my working hypothesis is that this kind of distribution, with evidently two peaks, centered, in this histogram at -20F and +10F, and the long cold tail is the outcome of two distinct temperature regimes: clouds vs. no-significant clouds.  

Lastly, just glancing at the first plot above, the recent mild New Year's Eve temperatures not withstanding, there does not appear to a strong trend in the temperatures, though there are notably fewer years at 20 below or lower since the early 1980s. In fact, there is something of trend toward warmer New Year's Eve temperatures. The linear regression (green line below), which is effectively modeling the mean value and is sensitive to outliers, does not meet the threshold for significance (i.e. we can not say with reasonably high confidence the apparent increase is not due to simple random variation). However, if we are concerned with the median New Year's Eve temperature (not sensitive to outliers), then the trend of 21F warmer now than in the early 1940s does meet a statistical significance threshold. 
So what will the upcoming New Year's Eve bring? Well, the airmass is likely to be pretty mild, and there is reasonable agreement that there will be a broken to overcast cloud cover, so I'll go with a comparatively mild ringing-in of 2018.  Happy New Year to you and yours, and thanks for reading Deep Cold. --Rick T.

Satellite Temperatures

Seasonably cold weather has returned at last over interior Alaska, with Fairbanks seeing generally sub-zero temperatures in the past week and a chilly high temperature so far today of only -19°F.  The cooperative observer at Chicken reported temperatures below -40° the past three nights.

The map below shows a satellite-derived estimate of temperatures at 4am this morning over the southeastern interior, extending from Fairbanks-land (near the top left) to the Canadian border and western Yukon Territory on the right.  The cold conditions of the upper Tanana River valley and the steep valleys of the Fortymile Country and western Yukon are clearly evident, with estimated temperatures around -35°F to -45°F.  Ground truth observations provide confirmation: -38°F in Northway this morning, and below -40° in Dawson, but barely below 0°F in Eagle with a breeze blowing.

In recent months I've been working on a project to decode all of the wintertime historical land surface temperature data for Alaska from the MODIS instruments on the Aqua and Terra polar-orbiting satellites.  The map above is derived from the realtime MODIS feed, but the entire archive is available from NASA (back to 2000 for Terra and 2002 for Aqua).  The full archive is of course a lot of data and takes a long time for my little computer to decode, so I'm only about half way through; but the preliminary results are already quite interesting.

As an example of the kind of analysis that is possible with the data archive, the maps below show the number of days with estimated surface temperatures of -50°F or lower for several winters (November through March).  The numbers are conservative estimates, because I've used only "high quality" pixels and there may well have been times when ice fog, cloud cover, or other problems prevented good estimates; but the temperatures are unlikely to be too low and so the counts of days are probably not too high.

Click the images to enlarge.  The Old Crow Flats of northern Yukon stand out as remarkably and consistently cold.  The results also support Rick's opinion that there are places upriver from Umiat that are quite a bit colder than that notoriously cold place.

Nov 2002 - March 2003:

Nov 2003 - March 2004:

Nov 2004 - March 2005:

Nov 2005 - March 2006:

Nov 2006 - March 2007:

Alaska Range Snowfall

I don't often comment on the weather and climate of southern Alaska, but this week a very interesting paper was published looking at long-term snowfall accumulations high on Mt Hunter in the Alaska Range.  Here's a summary from the Weather Company

https://weather.com/news/climate/news/2017-12-19-warming-ocean-water-double-snowfall-alaska-study

and here's the study itself

https://www.nature.com/articles/s41598-017-18022-5

The paper begins by reminding us of the dramatic increase in winter precipitation that has occurred in recent decades at coastal sites in southern Alaska, for example (and most strikingly) a more than 50% increase in rain and snow at Kodiak.  The chart below shows the December-February precipitation totals since 1950 at Kodiak; note that the increase essentially occurred as a step change in 1976 when the PDO suddenly flipped from negative to positive.  The median December-February precipitation changed from 13.9" (1950-1975) to 23.6" (1976-2016), an increase of 69%, but there has been no significant trend since 1976.

The precipitation estimates from Mt Hunter, as published in the new study, show a very dramatic increase over the past 1000+ years, and they also indicate that snowfall has continued to increase in the past few decades.  Interestingly the 1976 Pacific regime shift is not particularly clear in the Mt Hunter data (see below); a more pronounced increase occurred after 1985, and the highest estimated totals occurred in the last decade of the data (which ends with the winter of 2010-2011).

The absence of a 1976 step change at Mt Hunter raises a question as to how closely the precipitation amounts are connected to the North Pacific circulation pattern.  The authors of the study argue that the long-term strengthening of the Aleutian Low is the main driver for the long-term increase in Mt Hunter snowfall, so this is an important point.

The chart below shows two measures of North Pacific atmospheric variability that are referenced in the paper: the Pacific-North American (PNA) index and a North Pacific (NP) pressure index; both of these are closely related to the strength of the Aleutian Low.  The usual definition of the NP index means that it's more negative when the Aleutian Low is stronger, but I've defined it with an inverse sign to line up with the PNA.  The 1976 regime change is evident for both indices, but as with the Kodiak precipitation data there has been no significant increase since then - in fact the PNA index has zero trend since 1976 and the NP index has a negative trend.

A scatter plot of the PNA index versus Mt Hunter precipitation (see below) reveals a fairly robust connection with - as it turns out - the same correlation for 1950-1975 and 1976-2010.  The correlations are slightly lower for the NP index (+0.39 in both periods).

It's clear from these results that the Mt Hunter data does reflect an influence of the Aleutian Low, but it's equally clear that we can't explain the most recent (post-1985) increase in snowfall by invoking continued strengthening of the Aleutian Low.  The published study alludes to this point and suggests that the explanation for the most recent changes may be a sustained increasing trend in ocean temperatures far away in the western tropical Pacific Ocean; these tropical changes may have caused shifts in the atmospheric flow near Alaska that are not reflected by the PNA/NP indices and/or the sea-level coastal precipitation measurements.  The authors describe this hypothesis as "a heightened sensitivity to tropical SST teleconnections at higher elevation in Alaska".

While I don't have any particular comments on the tropical-connection hypothesis, I thought it would be interesting to look "closer to home" for potential explanations for the most recent increase in Mt Hunter snowfall.  To begin, I calculated the portion of the Mt Hunter precipitation that cannot be explained by a linear regression with the PNA index - see below.  The most obvious aspect of the chart is the preponderance of high precipitation totals, and the absence of small totals, since 1988-89.

It's not immediately obvious what major climate regime shift could have caused a change like this in 1988; as we noted before, the PDO shifted in 1976, and the Atlantic Ocean shifted into a much warmer state (positive AMO phase) in 1995; but I'm not aware of any notable shifts in or around 1988.

To dig in a little deeper, I calculated the correlation between September-April sea-level pressure and the Mt Hunter precipitation residual.  The results show an interesting signal related to pressure over the Chukchi Sea - see below.  When low pressure prevails over the Chukchi Sea, then Mt Hunter snowfall tends to be higher than the PNA would suggest, whereas high pressure to the north and northwest of Alaska tends to occur with lower Mt Hunter snowfall amounts.

Looking at an area-average of sea-level pressure over a box from 65-80°N and 160-180°W, we see an apparent shift to lower values beginning in the late 1980s (see below), so this lines up rather well with the unexplained recent increase in Mt Hunter snowfall.  A scatter plot of the annual pressure and precipitation residual values (second chart below) reveals a 1976-2010 correlation of -0.48, which is nearly as good as the underlying PNA correlation that we started with.

Of course correlation does not prove causation, but there does seem to be some evidence here that lower pressure to the north and northwest of Alaska may have contributed to the continued rise in Mt Hunter snowfall in recent decades.  How would this work from a physical standpoint?  Lower pressure to the north would seem to favor episodes of strong westerly flow across all of Alaska, and while southerly flow is most favorable for heavy snow in the Alaska Range, it's conceivable that westerly wind events could also transport ample moisture into the high terrain.  Further investigation is clearly required to see if there is really a plausible physical connection to storminess in the Chukchi Sea.

In summary, the remarkable hydrological changes that have occurred recently on Mt Hunter and presumably elsewhere in the Alaska Range appear to have some connection to Arctic weather patterns as well as to the North Pacific circulation.  This raises the interesting possibility that the recent reduction in Arctic sea ice may also be implicated, perhaps by altering the winter flow patterns around Alaska in a systematic way.

Yet More Warmth

Blog updates this month are starting to sound like a broken record, but that's also an apt description of the extraordinary weather pattern that continues to pump warm air northward across Alaska.  The warmth has been very anomalous, but the persistence of the pattern may be the biggest story, as Fairbanks has now seen 17 consecutive days with a daily high temperature of 20°F or higher and a low temperature above 0°F.  The 1981-2010 normals for the time of year are around +5°F and -12°F for the high and low respectively.

Looking back at the Fairbanks history from 1930-present, the December-February record for consecutive days above 0°F is 18 days, so it looks like this may be broken soon.  It's a pretty amazing achievement when you consider that any given day is about 80% likely to see 0°F or lower in December in Fairbanks (all else being equal, i.e. based on the 1981-present climatology).

As I mentioned the other day, the 3 long-term climate sites of Fairbanks, McGrath, and Bettles have never recorded such a lengthy period in winter with a combined average temperature of more than 1 standard deviation above normal.  Of the 3 locations, Bettles has been the warmest relative to normal, and they haven't even seen a sub-zero temperature so far this month (see figures below).  No other year in their climate record comes close to the warmth for the first 18 days of December, although of course Bettles climate data does not include December 1934 - and that remarkable month was marginally warmer than this year in Fairbanks through the 18th (but then turned colder in the last 10 days of the month).

Warm Air Aloft

In pondering the nature of winter warm-ups over Alaska, I find it interesting to consider how much of the atmospheric column rises above freezing, even while surface temperatures usually remain below freezing, at least for most interior locations.  In the most recent case, for example, the Thursday afternoon sounding from Fairbanks reported that nearly 20% of the atmosphere (by mass) was above freezing above the Golden Heart City.  The deep layer of above-freezing air is evident in the diagram below where the red line (the temperature trace) extends to the right of the diagonal blue 0°C line from the surface up to about 800 mb.

If we look at historical Fairbanks sounding data back to 1958, we find that the deepest layer of above-freezing air observed in December was in 1985, when the freezing level was way up around 8000 feet above ground (above 750 mb) on the 11th of the month.  The most extreme case to occur at any time in winter was the amazing warm spell of late January 2014, as documented here.  In that case over 30% of the atmosphere by mass was above freezing, and that's higher than normal for the height of summer.

The chart below shows a box-and-whisker depiction of the history of monthly means of the above-freezing fraction.  The numbers are calculated by adding up the pressure depth of each above-freezing layer in each sounding, after interpolating the freezing level(s) between reported heights.  Note that the record values labeled on the chart are for the monthly means, not for individual events.  It's interesting to note the sustained extreme warmth observed in some recent years in spring and summer, but not so much in winter.

A quick look at trends by month reveals that April is easily the month with the strongest warming trend based on this metric; the trend is actually quite remarkable (see below).  The mean value for June-August also shows a statistically significant upward trend, but I'll refrain from making a similar statement about the winter trend as a highly non-Gaussian distribution makes the statistics a bit more tricky.

Exceptional Warmth

Unusual warmth is prevailing across Alaska with remarkable persistence so far this month, and yesterday was the warmest day yet for the interior; McGrath reached 39°F, Fairbanks made it to 41°F, and Bettles (at nearly 67°N latitude) hit 38°F.  The Bettles high temperature tied the record for the month of December, with data back to 1951.

Looking at the joint history of climate data from these three interior sites (1951-present), there is only a handful of days when all 3 locations reached 38°F or higher on the same day in the depths of winter (December through February).  It has happened only 6 other times, with the most recent occurrence being nearly 8 years ago (Jan 17, 2009).  January 2009 was also the last time that the daily mean temperature was at least 2 standard deviations above normal for all 3 of these sites on the same day in winter (November-March).

With respect to persistence, we are currently at 11 straight days with a combined mean daily temperature (for the 3 sites) of more than 1 standard deviation above normal.  The record for November-March is 13 days, so by this metric we're closing in on a record warm spell for the interior.

The chart below shows the kind of warmth that's been observed by the soundings from Fairbanks; for the first half of this week, temperatures were around 40°F or higher at about 600' above the surface.

Here's the 500mb height analysis from 3pm yesterday.  If it looks oddly similar to the chart from last week... that's because it is.

Warm Days in Winter

Warm chinook flow from the south lifted temperatures above the freezing point on Thursday night in Fairbanks for the first time in more than a month.  Thursday's high temperature, which occurred at 10pm, was 37°F, but this isn't close to a daily record; the extraordinary chinook of early December 1934 set all the high temperature records in Fairbanks from the 3rd to the 8th of the month (including 58°F on December 5).  Nevertheless, the flow anomaly from this week's event was impressive; the map below shows the 500mb height analysis from 3pm Thursday.

It's obviously unusual for temperatures to rise above freezing in the depths of winter in Fairbanks, but most years do see at least a brief thaw at some point in December through February; only 10 years since 1930 remained below freezing throughout the 3-month period (most recently 1998).

The chart below shows the number of days each winter on which the daily high temperature was above 32°F in Fairbanks, and the red shading indicates how many days the daily mean temperature was also above freezing - the latter being a much rarer event.  The relative abundance of warm days prior to about 1945 may have something to do with the fact that measurements were taken downtown from 1930-1942, rather than at the international airport as they have been since 1951.

The asterisk annotations indicate winters when La Niña conditions were present in the tropical Pacific Ocean, as they are this year.  Despite the fact that La Niña winters tend to be colder on average in Fairbanks, the frequency of warm days is not much reduced, and this is because the temperature variability is heightened during La Niña winters.

The scatter plot below shows how the number of above-freezing days varies with an index of the El Niño/La Niña (ENSO) phenomenon.  A clear peak is evident for near-neutral ENSO conditions, and this is an interesting example of non-linearity in climate behavior; while El Niño winters tend to be warm on average in Fairbanks, the number of very warm days is no higher than normal, because temperature variability is low.  I've discussed this kind of thing before.

The same chart for the PDO rather than ENSO shows a hint of the same behavior, although there is more asymmetry between the positive and negative phases.  There is a notable reduction in warm day frequency when the PDO is significantly negative, but so far this winter the PDO is near-neutral.

Warm North, Cold Southeast

November brought a sharp contrast in temperature anomalies across the state of Alaska, with extreme warmth (relative to normal) in the north and below-normal temperatures in the southeast interior, south-central, and southeast Alaska.  Here's Rick Thoman's graphic from the other day:

Alaska is a big place and often has significant regional climate differences. November 2017 temperature departures are no exception. Utqiaġvik warmest Nov, St. Paul 2nd warmest, Juneau coolest Nov since 2006. #akwx #Arctic @Climatologist49 @DaveSnider @AK_OK pic.twitter.com/WnL7os4sWM

— Rick Thoman (@AlaskaWx) December 2, 2017

As an example of the extremes, Utqiaġvik (Barrow) saw their warmest November on record, but the notorious cold spot of Chicken reached -30°F on 12 consecutive days, which is the most on record prior to mid-late December (but the period of record only goes back to 1997).  The chart below shows how the temperatures at the two sites diverged in November; note that Utqiaġvik had several days of missing data.

As it turned out this year, Utqiaġvik had "only" their 8th warmest October since 1930, but the warmth in November was unprecedented at 16°F above the 1981-2010 normal.  The early November peak in anomalous warmth is completely consistent with the absence of sea ice nearby; the chart below (taken from this 2014 post) shows an early November peak in the recent warming of surface air relative to air aloft at Utqiaġvik.

As for the weather pattern that led to the statewide contrast last month, we can blame La Niña for a persistent ridge of high pressure that allowed cold conditions to prevail downstream over southeast Alaska.  The first map below shows the November mean 500mb height - note the ridge over the Bering Sea - and the second map shows the departure from normal.  This is a classic La Niña pattern; but the warmth in the far north has more to do with the sea ice situation.

Sea Ice Update

Freeze-up is progressing at an extremely slow pace this year in the Arctic and near-Arctic ocean waters surrounding Alaska.  The latest sea ice analysis from the National Weather Service continues to show a very large region of open water in the Chukchi Sea, and it's only in the past couple of days that the coastal water has frozen up at Utqiaġvik (Barrow).

The absence of ice in the Chukchi and Bering Seas is unprecedented for the time of year in the period of record since satellite monitoring began.  According to the NSIDC's regional sea ice index, the Chukchi Sea's ice extent was 25% below the previous record (in 2006) as of yesterday.  Another way of looking at it is that yesterday's ice extent was equal to the 1981-2010 normal for October 13; so in comparison to that reference period, we could say that freeze-up is running about 7 weeks late.

Here's a chart of the historical November average ice extent for the Chukchi and Bering Seas.  This year's November average of 320,000 km² is very nearly the same as the 1981-2010 normal for September, when seasonal melt-out reaches its peak.

And here's a simple animation of the last 10 days of ice analysis.  From 1979-present, the Chukchi Sea has never failed to freeze up completely before the end of the year (last year it got there on December 30); but this year it still has a long way to go.